Strange behaviour when multiple active queries about pq

[fdr]

Some of these problems can be consequences of design choices in database/sql:

http://golang.org/src/pkg/database/sql/sql.go#L176

The fact that you are using db.Query twice in a row before closing the two row sets at the end is probably causing something weird.

Can you submit a test case for the version that blocks? That feels like a definite bug.

[bgentry]

The fact that you are using db.Query twice in a row before closing the two row sets at the end is probably causing something weird.

That should just result in a new connection being allocated for the connection pool. It shouldn't cause any conflicts.

[fdr]

That should just result in a new connection being allocated for the connection pool. It shouldn't cause any conflicts.

That's my thinking, too. Hence, the request for this variant of the test.

[kisielk]

Okay, so if I understand correctly, the reason I'm getting the "relation temp2 does not exist" error is because the sql package starts up a new connection which now doesn't haver this temp2 table. That makes sense.

Unfortunately I don't have the test case with the transaction handy right now since I just fixed the code and moved on (it was a fair bit more complicated than the provided example, so I was trying to simplify), but I will make an attempt to reproduce it this weekend and get back to you. Thanks guys.

[kisielk]

I just found this in the Go bug tracker today: https://code.google.com/p/go/issues/detail?id=3857

Sounds like what I was experiencing, so it looks like it's more like a bug in database/sql than specific to this driver.

[fdr]

Thanks for investigating. Yes, I believe you are right. Unfortunate. Re-open if you think this bug has more to it on the pq side.